Gait Recognition using RGB-D Camera for Player Labeling

The identification of the 3-dimensional physical properties of humans has recently become a very important field of study. In this respect, today's applications and research use depth cameras such as Kinect as an input tool for identification of quantitative characteristics of people. Due to the completeness of data offered by conventional cameras, depth-cameras are the driving force for more up-to-date research. Introduction of human objects recognition offered by its library makes researchers easily take three-dimensional data from Kinect for skeletonization algorithms. The skeletonization is an algorithm that can facilitate the depth-camera’s user to detect human objects and track the bone and each joint. In this way, a human can be tracked precisely by bone and joint motion. Moreover, the camera is able to do skeleton tracking over a human object. However, if the human object is actively moving in the field, the skeleton mapping may change. In this paper, we propose a new method for labeling players by using gait recognition and other features thus even the human is moving in/out of the camera field, the camera will recognize it as the same human based on its characteristic. This approach can be easily used in real-time recognition of people and is also performed to label the player. Experiments of this method were carried out in the real environment. Tests were performed successfully in experiments where people with different characteristics are involved. According to the obtained results, the proposed method has a good performance and can be applied in real-time easily.